1. Identification
| Item | Detail |
|---|---|
| Instrument | Presidential Circular (2026/9) |
| Official Gazette | 18 August 2026 — issue 33344 |
| Signed | 17 August 2026 |
| Subject | Türkiye AI Action Plan (2026-2030) |
| Coordination | Ministry of Industry and Technology |
| Monitoring | National AI Council |
| Predecessor | National AI Strategy (2021-2025) |
2. Normative character: the limits of bindingness
A circular is an administrative instrument addressed to public institutions. Being neither a statute nor a regulation, it imposes no directly enforceable obligation on private parties. In practice, however, it will reach the private sector through three channels:
- Public procurement: the minimum 2% share of public investment earmarked for AI, and the public-sector adoption target, will crystallise as technical and compliance criteria in tender specifications.
- Secondary legislation: regulatory sandboxes and data-sharing commitments could well be turned into binding rules through regulations and communiqués.
- Supervisory expectation: monitoring through the Council will establish a de facto compliance standard for suppliers working with the state.
3. The four axes
| Axis | Focus |
|---|---|
| Discover | Public awareness and AI literacy |
| Benefit | Adoption across public and private sectors |
| Produce | Domestic technology, models and infrastructure |
| Govern | Governance, regulation, oversight and ethics |
A total of 16 actions sit beneath these axes.
4. Quantified targets
| Target | Scale / horizon |
|---|---|
| Installed data-centre capacity | 1 GW (2030) |
| Infrastructure investment mobilised | at least USD 10 billion |
| Share of public investment for AI | at least 2% |
| Advanced specialists | 10,000 |
| Application professionals | 100,000 |
| AI literacy training | 5,000,000 citizens (2 years), 81 provinces |
| Open public datasets | 2,000+ (National Data Library, 2030) |
| Regulatory sandboxes | 5+ |
| Single-window approval | max. 30 business days |
5. Principles
The plan is built on human-centricity, reliability, ethical responsibility, digital sovereignty and sustainable development. While this set does not map one-to-one onto the risk-based architecture of the EU AI Act, it points in the same conceptual direction on human oversight and trustworthiness. For companies operating in Türkiye that also place products or outputs on the EU market, the elements the two frameworks share (documentation, data governance and human-oversight records) can be kept in a single compliance file.
Practical takeaway for companies
For now, the action plan creates no obligations; it is better read as a forecast of the compliance calendar. For organisations that work with the state or operate in regulated sectors, reasonable first steps are to inventory the AI systems in use, record data sources and their legal bases, put human-oversight thresholds in writing, and prepare the technical documentation likely to be requested in tenders.
